what a big thread... who started all this thing? ;-) On Wed, 1 Oct 2003 16:22:12 -0230, Juhan Leemet wrote / a écrit: > GuyCLO~ wrote: > >I agree. I found that larger latencies (100 < latency < 200) are usable for > >having fun. I mean I have used softsynths on a computer without tuning > >latency and without beeing root. > > Hmm. How are you measuring latency? I'm not sure how to do it (sufficiently > accurately). I have played with soundtracker. In the menu, you have the buffer choice, it gives you an estimated audio delay. I played samples with the keyboard and found that it was good for me (am I too easily satisfied?) . I was using a non-optimized mandrake kernel. Of course I said "for having fun". When things get more serious, switching to a low-latency kernel is needed. > I'm running SuSE 8.2, which some have suggested includes the > low-latency patch, but I don't think so. I've been too busy (lazy?) to check > or implement. It's easy to check in my redhat+planet ccrma distro: cat /proc/sys/kernel/lowlatency and the answer should be: 1 > I find even when playing some .ogg file to jam along with, that > hiccups are "disturbing" or distracting. Maybe I've got problems (as a > musician wannabe) with my timing? I've got my .ogg files on a server, but I > believe xmms pre-buffers (I recall setting it to 1/2 second at one point?) xmms's buffer shouldn't be a problem, it's just for monitoring. It's the apps (softsynth) that you play that should be responsive. If you don't use softsynth and you use xmms just for monitoring, hiccups through your network should lead to a network analysis. Maybe there are other net messages that take bandwith ? Anyway, if you want to play with audio apps like soundtrackers, softsynth, etc. I warmly recommend you to enable/install a low-latency + preempt patched kernel. The latency is dramatically reduced ( _largely_ under 40 ms ) and best of all, everything is more responsive ( Xfree! ). It can be a real pain to put up. I am pleased to have done it with my mandrake, but I prefered to switch to the redhat based Planet CCRMA packages. Installing the low-latency kernel is pretty quick and straight-forward with this meta-distro and the audio apps are rpm packaged. [snip] >You would have to transmit a time code in each > packet, so the player can continually "re-sync" the audio. I don't think the > current audio streams do that? I think they "assume" (Benny Hill?) that the > audio stream is continuous, and therefore you can derive the timing? as pointed in another post, I was thinking about MIDI only packets, not audio ones... sorry that I wasn't clear about it. Of course, on the target machines, low-latency should be enabled, because the quicker the MIDI flux will be played through the soundcard, the best. greetings GuyCLO~ ----------------------------------------------- #219055 http://counter.li.org ----------------------------------------------- http://perso.wanadoo.fr/guy.clotilde/index.html ----------------------------------------------- Change your Linux PC in a Rythm Station with holborn's GMORGAN: http://perso.wanadoo.fr/guy.clotilde/GMORGAN/index.html ----------------------------------------------- Quidquid latine dictum sit, altum viditur. (Whatever is said in Latin sounds profound.)